alpha is simply a parameter, typically the loop gain (but it can be
any parameter). Let us call it K for convenience. The aim of the root locus
method is to plot the roots of 1+KPC (i.e, solutions to 1+KPC=0) as K varies.
Rewrite 1+KPC=0 as PC=-1/K.
When K=0, we see that we need PC to be
"infinite" for this equation to be satisfied. This happens at the poles
of PC. Thus when K=0, the roots of 1+KPC are at the poles of PC.
When K goes to infinity, we need PC to be zero for the equation to be satisfied.
This can happen in two ways -(a) if the roots are at the zeros of PC or (b) the
roots approach "asymptotes" defined appropriately (CDS 110).
For now, just consider case (a). So when K goes to infinity, we
have that the roots go to the zeros of PC.
Thus the statement that "the roots of 1+KPC go from the poles of PC to the
zeros of PC" (it is understood in this context that we are increasing K from zero to infinity).